{"id":"W4360879252","doi":"10.2196/46348","title":"Deep Learning Approach for Negation and Speculation Detection for Automated Important Finding Flagging and Extraction in Radiology Report: Internal Validation and Technique Comparison Study","year":2023,"lang":"en","type":"article","venue":"JMIR Medical Informatics","topic":"Radiology practices and education","field":"Medicine","cited_by":9,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Flagging; Computer science; Artificial intelligence; Negation; F1 score; Natural language processing; Encoder; Transformer; Security token; Speculation; Machine learning; Programming language","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002011468,0.0001015916,0.0002460496,0.0003213045,0.0001448008,0.00003930039,0.00002178992,0.0002123649,0.000002051351],"category_scores_gemma":[0.001128943,0.00009299912,0.00001863369,0.0001857919,0.00004644598,0.000372706,0.00002054961,0.0002802571,2.568404e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007728917,"about_ca_system_score_gemma":0.00003455008,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002349558,"about_ca_topic_score_gemma":0.00001339135,"domain_scores_codex":[0.9987372,0.00006363348,0.0007361444,0.0001482827,0.0001544149,0.0001603974],"domain_scores_gemma":[0.9989851,0.0003267955,0.0004719301,0.00006966756,0.0000611615,0.00008540766],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001075095,0.0004270386,0.7281489,0.001964183,0.0001726878,0.00002254305,0.03879019,0.002292445,0.006351869,0.0001267013,0.0001869681,0.2204414],"study_design_scores_gemma":[0.001231404,0.0004598702,0.1252276,0.00006013098,0.0000540196,0.0008707311,0.009696404,0.8618391,0.000268144,0.00005818694,0.0001569925,0.00007749954],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8237594,0.00002698372,0.1740364,0.0001559282,0.0001038913,0.001750323,3.766238e-7,0.00013979,0.00002682421],"genre_scores_gemma":[0.9885688,0.00008288927,0.01039384,0.00002337657,0.0001031133,0.0005263037,0.000275383,0.00001074208,0.00001556919],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8595466,"threshold_uncertainty_score":0.3792394,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04012883448843974,"score_gpt":0.400929795150139,"score_spread":0.3608009606616993,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}